Loki – A Lava Lake in Rarefied Circumplanetary...
Transcript of Loki – A Lava Lake in Rarefied Circumplanetary...
Loki – A Lava Lake in Rarefied
Circumplanetary Cross Flow
Andrew C. Walker, David B. Goldstein, Philip L. Varghese, Laurence M. Trafton,
and Chris H. Moore
University of Texas at Austin, Center for Aeromechanics Research, 1 University Station C0600, Austin TX, 78712
Abstract. The interaction between Io’s largest hot spot, Loki, and Io’s circumplanetary winds is simulated using the
direct simulation Monte Carlo (DSMC) method. Our three-dimensional simulation models the rarefied pressure-driven boundary layer flow over a “hot” disk in the presence of a weak gravitational field. The pressure gradient which forces winds away from the subsolar point toward the nightside is caused by the variation in insolation over the surface. The rarefaction varies strongly with time of day due to the exponential dependence of the vapor pressure on the surrounding
surface frost temperature (KnHS ≈ 1×10–4 to 0.5 where KnHS = λ/R, λ is the mean free path, and R is Loki’s effective
radius). The spread of heat from the hot spot, the equilibration of pressure over the hot spot, and separation of the
boundary layer are examined. The spread of heat away from the hot spot is approximately controlled by δ = tRADU/R
(tRAD is the radiation time scale and U is the mean wind speed). For cross flow speed considered here, δ ≈ 0.5 and
therefore the gas warmed by the hot spot cools by e–1 ~ 0.5R downstream of the hot spot edge. For the cases without plasma heating, the boundary layer flow separates near the hot spot because the spot creates a significant adverse pressure gradient. Despite the near surface pressure over the hot spot being lower than over surrounding regions, the increased scale height due to the 332 K surface temperature results in higher pressures above the hot spot than the surrounding
sublimation atmosphere at high altitudes (>10 km). When plasma heating from above is included the atmosphere is significantly inflated leading to a higher pressure gradient at all altitudes and therefore higher flow speeds. The elevated pressure at high altitudes also decreases the relative size of the adverse pressure gradient created by the hot spot; therefore the boundary layer remains attached. The pressure over the hot spot does not equilibrate with the surrounding sublimation atmosphere because dB-ATM < R and dB-HS > R (where dB-HS is the ballistic length scale over the hot spot); i.e. SO2 molecules from the local sublimation atmosphere penetrate only dB-ATM/R into the hot spot and the majority of those molecules will hop back outside the hot spot in a single ballistic trajectory.
Keywords: Io, Atmospheric dynamics. PACS: 96.12.Jt, 96.30.L-, 96.30.Ib
INTRODUCTION
Io, the innermost Jovian satellite, is the most volcanically active body in the solar system due to tidal flexing
resulting from its orbital resonance with Europa and Ganymede. Lava lakes, lava flows, and warm calderas (often
called hot spots) are common features on the surface due to this tidal flexing. Many of these features are also sources
for volcanic plumes. One of the most prominent hot spots is the lava lake Loki. Observations have determined that
the Loki volcanic region accounts for between 30% [1] and 36% [2] of the total heat flow from Io. Marchis et al. [2]
used the Keck telescope to observe Loki and calculate an effective temperature of 325-340 K and an effective area
of 10.3-11.5×103 km2. Io has also been observed to have a rarefied atmosphere in which the dominant dayside species is SO2. The two
primary sources of the atmosphere are volcanic plumes and sublimating SO2 surface frosts; however, the dominant
source has not yet been determined due to conflicting observations. Pressure gradients in the sublimation
atmosphere (due to the variation in insolation) drive circumplanetary flow which diverges in all directions away
from the peak pressure region near the subsolar point (the point nearest the sun) toward the nightside.
Io’s rarefied atmosphere has been modeled many times before; the first model was proposed in 1985. Ingersoll et
al. [3] vertically integrated and solved the conservation equations to find sublimation-driven flow. In 1989, Ingersoll
modified the earlier work to account for the non-uniform surface properties [4]. The key idea was the introduction of
a horizontal averaging length scale over which any surface frost patches would locally control the surrounding
atmospheric pressure. Later models have included radiative transfer, plasma heating, photo- and neutral chemistry
[5-9]. All of these simulations modeled Io’s atmosphere as a continuum; this is a poor approximation in rarefied
conditions (near the nightside and at high altitudes) and a rarefied gas dynamics method such as DSMC must be applied. More recently, Io’s atmosphere has been modeled using the DSMC method [10-13].
In the present paper, we examine the rarefied pressure-driven boundary layer flow of the atmosphere over a
“large” but finite hot region (Loki) and expand on the previous work of Walker et al. [12] which modeled only the
sublimation-driven flow without hot spots.
MODEL
To properly account for physics that occurs on Io’s surface and in its atmosphere, we have made significant
changes and additions to the original DSMC code developed by Bird [14]. These improvements have been documented in detail in earlier work by Walker et al. [12]. A brief summary of the additions follows.
SO2 molecules undergo a gravitational body force consistent with Io’s gravity causing them to follow ballistic
trajectories after they are sublimated from surface frosts. The SO2 rotational energy states are treated as continuous
while the vibrational energy states are treated as discrete. These energy states are initially populated by assuming
that sublimated molecules are in thermal equilibrium with the surface frost. Energy exchange between translation,
rotation, and vibration can occur during collisions. Radiation from the internal energy states also occurs via
spontaneous emission but all photons are lost since the atmosphere is assumed optically thin. Plasma bombardment
by ions and electrons (from the Jovian plasma torus) is modeled via an energy flux. The energy flux is initialized to
1.3 erg/(cm2s) [15] at the top of each column of cells, travels radially downward, and is absorbed by gas in each cell
dependent on its SO2 density. The energy absorbed by the gas goes into rotation and translation while excitation of
vibrational modes is neglected. A more physically accurate model for plasma bombardment (which tracks ions and electrons) has been developed and is being tested for use in the three-dimensional DSMC code [16].
Hot Spots & Sublimation Atmosphere
Loki is approximated as a circular disk with an area of 10.8×103 km2 and a uniform surface temperature of
332 K. Loki’s actual location on Io has been altered to allow for easier computation; here Loki is located on the
equator at 45° W longitude (Loki’s actual location is 10° N, 310° W). The hot spot disk is assumed to not sublimate
since SO2 surface frosts would rapidly be burned off by the warm surface temperatures. Some “hot spots” are also
known volcanic plumes and have a significant mass flux; we assume that the mass flux from Loki is zero since it is known to be a periodic volcano with quiescent intervals. SO2 molecules which impact the disk of the hot spot are
desorbed during the same time step since the residence time at 332 K is on the order of 10 µs (much smaller than the
simulation time step of 0.5 s). Molecules that impact the hot spot surface desorb in thermal equilibrium with the
surface. The level of rarefaction over the hot spot can be characterized by defining a “hot spot” Knudsen number
(KnHS = λ/R) where λ is the mean free path in the local surrounding sublimation atmosphere and R is the effective radius of the “hot spot”.
One characteristic length scale of the sublimation atmosphere is the scale height, H = kT/Mg. Here k, T, M, and g
are Boltzmann’s constant, translational gas temperature, molecular mass, and gravitational acceleration,
respectively. An atmospheric Knudsen number can then be defined; KnATM = λ/H. For typical surface frost temperatures, the scale height near the surface varies between 8.3 km at 115 K and 6.5 km at 90 K. For convenience
and clarity, the rarefaction will generally be defined by KnHS rather than KnATM (with KnHS a factor of 6 to 8 smaller
than KnATM at low altitudes).
Simulation Parameters
The DSMC simulations are computed in parallel (with domain decomposition only in azimuth) using a fully
three-dimensional spherical grid. The domain covers longitudes between 0° W and 90° W, latitudes between 5° S
and 5° N, and up to 400 km in altitude. The lateral (azimuthal and polar) cell size is 1.6 km (1800×200 cells) while
the radial cell size grows exponentially with altitude to resolve the mean free path. The radial cell size is ~5 m near
the surface but is limited to 5 km at high altitudes. There are 190 radial cells in total. The domain is decomposed in
longitude across 360 processors. The boundary condition on the top surface of the domain is a vacuum boundary
condition as a negligible number of molecules reach this altitude and are lost. The boundary conditions at 5° S and
5° N are specular reflection. The boundary condition at 0° W located near the peak pressure region is a specular
boundary while the boundary condition at 90° W is a vacuum (since the nightside atmosphere there is essentially
collisionless; KnATM > 1). The lower surface boundary condition is more sophisticated than the other surfaces and is
explained in detail in the next section. The constant latitude size domain closely match the global wind speeds
computed in Walker et al. [12]; however, the actual global winds likely diverge more than the specular boundaries
currently allow resulting in some minor non-conformities in the present simulations [12].
FIGURE 1. The global domain of Io’s sublimation atmosphere with a cross-hatched area showing the reduced domain for these circumplanetary flow/hot spot interaction simulations. Loki is visible in the center of the sub-domain (which shows contours of the number density in the atmosphere) and an image of the lava lake taken from Galileo is also shown (the dark area is Loki).
Bottom Surface Boundary Conditions
Io’s surface is broken into two separate components: frost and rock. The frost fraction, 2,
SOf [17] determines the
percentage of frost (and rock) in any given surface cell. The frost and rock surfaces have independent temperatures
which differ due to their different albedos and thermal inertias. The number flux of sublimated SO2 molecules [(m2s)-1] is a function of the surface frost temperature and the surface frost fraction:
2
2
451013
1.516 10.
2
frostT
SUB SO
B frost SO
eN f
k T Mπ
−
×= (1)
In equation (1) the numerator is the SO2 vapor pressure [Pa] [18]. The surface frost and rock temperatures are then
determined by solving the one-dimensional heat conduction equation with their respective thermal parameters:
( , , , ) ( , , , )
,T z t T z t
c kt z z
θ φ θ φρ
∂ ∂ ∂ = ∂ ∂ ∂
(2)
where ρ is the density, c is the specific heat, k is the thermal conductivity, t is time, z is the depth into the surface, θ
is the latitude, φ is the longitude, and T is the surface temperature (rock or frost). The parameters ρ, c, and k can be
grouped together to form a single thermal parameter known as the thermal inertia, Γ. The thermal inertia values for the frost and rock are 120 and 30 ergs/cm2Ks1/2, respectively based on a best fit to the temperature distribution
observed by the Galileo PPR instrument [1]. Five constraints were used to find the best fit to observational data: the mean albedo, the peak brightness temperature, the peak frost temperature, rock thermal inertias lower than frost, and
rock albedos lower than the frost. The boundary conditions at the upper surface (z = 0) and at a depth where there
are no diurnal changes (z = d) are given, respectively, by:
( )4
0
( , , , )( , , 0, ) (1 ) ( , , ) ( , ) ( , , ) ( , , ),
S J SUB COND
z
T z tk T z t F t F q t q t
z
θ φεσ θ φ α θ φ θ φ θ φ θ φ
=
∂= = − − + − +
∂ (3a)
( , , , )
,T
z d
T z tk Q
t
θ φ
=
∂=
∂ (3b)
where ε is the emissivity, σ is the Stefan-Boltzmann constant, α is the albedo, FS is the solar flux, FJ is the reflected flux of sunlight from Jupiter, qSUB and qCOND are the heating terms due to the latent heat of sublimation and
condensation, respectively, and QT is the endogenic heating. The emissivity is assumed to be unity [1]. The frost and
rock albedos are based on the same best fit as the thermal inertias; the albedo values are 0.65 and 0.44, respectively.
Modeling of Galileo PPR observations showed that QT = 1.0 W/m2 best fits the data [1].
The frost fraction is also used to determine the probability that a SO2 molecule will hit frost or rock when it
impacts the surface. When molecules impact the frost surface they are deleted because the vapor pressure model
assumes a unit sticking coefficient; however, when molecules impact the rock surface they stick for a residence time
dependent on the surface rock temperature. The residence time is given by ( ) 0exp
RES S B rockt H k T ν= ∆ [19] where
∆HS is the surface binding energy of SO2 on a surface of its own frost (∆HS/kB = 3460 ± 40 K), and ν0 is the lattice
vibrational frequency (ν0 = 2.4×1012 s–1).
RESULTS
Three cross flow cases are investigated. Case 1 is the simplest case and includes homogeneous surface frosts
(uniform 50% rock / 50% frost surface) [17], unit sticking on the rock surface, and no plasma heating. Case 2
includes inhomogeneous surface frosts and a residence time on the rock surface but does not include plasma heating.
Case 3 includes inhomogeneous surface frosts, a residence time on the rock surface, and plasma heating.
x (km)
P(n
b)
dP
/dx
(pb
/km
)
TF
RO
ST
(K)
500 1000 1500 2000 25000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8 -8
-7
-6
-5
-4
-3
-2
-1
0 95
100
105
110
115
TFROST
(K)
dP/dx (pb/km)P
THEORY(K)
PACTUAL
(nb)
Loki
x (km)
Altitu
de
(km
)
0 500 1000 1500 2000 25000
10
20
30
40
50
60
KnHS
1.00E+00
5.08E-01
2.58E-01
1.31E-01
6.66E-02
3.38E-02
1.72E-02
8.73E-03
4.44E-03
2.25E-03
1.15E-03
5.82E-04
2.96E-04
1.50E-04
7.63E-05
3.87E-05
1.97E-05
1.00E-05
(A) (B)
Altitude (km)
n(m
-3)
Win
dS
pe
ed
U(m
/s)
Mo
me
ntu
mF
lux
(kg
/m2s)
0 10 20 300
2E+15
4E+15
6E+15
8E+15
1E+16
1.2E+16
0
50
100
150
200
0.0E+00
5.0E-09
1.0E-08
1.5E-08
2.0E-08
2.5E-08
3.0E-08
3.5E-08C2 nC3 n
C2 UC3 U
C2 M. FluxC3 M. Flux
x (km)
P(p
ba
r)
TT
RA
NS
(K)
n(m
-3)
1000 20000
10
20
30
40
50
60
70
80
90
25
50
75
100
125
150
175
200
0
1E+15
2E+15
3E+15
4E+15
5E+15
C2 P
C3 PC2 T
C3 TC2 n
C3 n
(C) (D) FIGURE 2. (A) The surface frost temperature, pressure above the surface based only on vapor pressure equilibrium (PTHEORY), the actual pressure above the surface (PACTUAL), and pressure gradient (based on PTHEORY) as a function of x (the distance away from the peak pressure region) along the equator. The actual gas pressure above the surface tracks vapor pressure equilibrium everywhere but in the vicinity of Loki. (B) KnHS as a function of altitude and x along the equator. Both A and B are for Case 1. (C) The number density, x-direction wind speed, and x-direction momentum flux for Cases 2 and 3 as a function of altitude. These profiles are taken ~70 km upstream of the upstream edge of the hot spot such that they are outside of the vortex created in Case 2. The x-direction momentum flux is very similar for Case 2 and 3 below 10 km but Case 3 has a much higher x-direction
momentum flux above 10 km (~6× greater at 20 km). (D) The pressure, number density, and translational temperature at 20 km as a function of x for Cases 2 and 3. The pressure gradient is much larger in Case 3 than Case 2 resulting in much higher wind speeds and the boundary layer remaining attached.
For Case 1, the pressure gradient creates rarefied boundary layer flow over the “hot spot”. The surface
temperature varies from ~115 K to ~90 K and therefore the near surface rarefaction varies from KnHS ≈ 1×10-4 to 0.5
(see Figure 2B). At high altitudes KnHS >> 1. The pressure near the surface varies from ~0.7 nbar at peak pressure to
~1 pbar near the nightside. A favorable pressure gradient, dP/dx < 0, exists everywhere but in the region near Loki
and the flow accelerates monotonically away from the peak pressure region toward Loki. The hot spot surface is a diffuse reflector causing the net x-momentum of the flow to be lost as molecules impact the surface. Furthermore,
the mean free path and ballistic length scale are both smaller than the radius of the hot spot (λ < dB-ATM < R, where dB-ATM is the ballistic length scale in the local sublimation atmosphere). The ballistic length scale is defined by
8 /B ATM B
d k T mgπ− = where m is the molecular mass. Thus molecules undergo many collisions along their paths
over the hot spot, and the vast majority of molecules will not pass over Loki in a single ballistic trajectory (dB-ATM <
2R). This combination of parameters leads to a moderately collisional (KnHS ≈ 1×10–3) stagnant gas above Loki. The
stagnant gas above Loki creates a substantial upstream adverse pressure. Although, the near surface pressure above
Loki is lower than that of the surrounding sublimation atmosphere, the pressure becomes higher than the
surrounding atmosphere at ~10 km. As the atmospheric flow encounters the large adverse pressure gradient created locally by Loki, the boundary layer separates and a vortex forms (see Figures 3A and 3B). Because of the high x-
direction momentum flux of the winds there is significant downstream deflection of the hot gas above the hot spot
on the upstream edge. The deflection of the hot spot gas from the vertical is ~35°.
x (km)
Altitu
de
(km
)
1300 1350 1400 1450 1500 15500
20
40
60
250
208
173
144120
100
83
6958
48
40
TTRANS (K)
Loki
(A) Case 1 – Homogeneous 50:50 frost/rock, unit sticking on rock, no plasma heating
x (km)
Altitu
de
(km
)
1300 1350 1400 1450 1500 15500
20
40
60
250
208173
144
120
100
8369
58
48
40
TTRANS
(K)
Loki
(B) Case 2 – Inhomogeneous frost/rock distribution, residence time on rock, no plasma heating
x (km)
Altitu
de
(km
)
1300 1350 1400 1450 1500 15500
20
40
60
250
208173
144
120
100
8369
58
48
40
TTRANS (K)
Loki
(C) Case 3 – Inhomogeneous frost/rock distribution, residence time on rock, plasma heating (1.3 erg/cm2s)
FIGURE 3. Contours of translational temperature as a function of altitude and distance away from the peak pressure region along the equator. Streamtraces in white show the circumplanetary flow’s interaction with the hot spot, Loki. The rarefied
boundary layer flow which had developed separates due to the adverse pressure gradient created by the hot spot in Cases 1 and 2. In Cases 1 and 2, streamtraces continue downstream of the hot spot due to three-dimensional effects as the flow curves around the hot spot. In Case 3, the boundary layer flow remains attached.
The situation is largely the same for Case 2. Modeling the residence time on the rock surface slightly increases
the day-to-night pressure gradient leading to higher momentum flux in the lower atmosphere. This increases the
deflection angle to ~45° (see Figure 3B). For Case 3, the plasma heating significantly alters the dynamics. The
plasma inflates the atmosphere (i.e. higher number densities at higher altitudes); therefore, the atmosphere becomes
collisional at higher altitudes. The temperature is also increased by the absorption of plasma energy; however, the
effects vary with x because the plasma energy only penetrates a fixed column of gas. The resulting pressures at
higher altitudes can be see in Figure 2D. The pressure gradient in Case 3 is much larger than Case 2 and this causes
higher flow speeds at higher altitudes (and a higher x-direction momentum flux; see Figure 2C). At high altitudes, the relative size of the adverse pressure gradient created by the hot spot is reduced by the higher pressures due to
plasma heating. The pressure gradient remains largely favorable even over Loki and the boundary layer remains
attached (see Figures 2D and 3C).
For Cases 1 and 2, the hot gas above Loki forms a wake of warm gas because of the comparable radiation and
convection time scales of the SO2 molecules. A fraction of the molecules that impact the hot spot (and subsequently
desorb in thermal equilibrium) will convect downstream with the circumplanetary flow; since KnHS ≈ 1×10-3,
collisions will transfer translational energy to the rotational and vibrational energy modes. The energy transferred to
vibrational modes is radiated rapidly because the largest Einstein A coefficients of the three SO2 vibrational modes is
0.88 s–1; therefore, each the vibrational mode radiates most of its energy before the SO2 molecule travels far from
the hot spot (tVIBU/R << 1, where tVIB is the vibrational radiation time scale). However, the time scale for radiation
from the rotational energy states of SO2, tROT, is on the order of 250 s at ~300 K (tROTU/R ≈ 0.5). The warm gas from
Loki cools by e-1 after the gas convects ~25 km (or ~0.5R) downstream of the hot spot edge (see Figures 3A and 3B). However, this simplifies the situation a bit as the gas must be collisional to transfer energy away from
translation and into the rotational and vibrational internal energy states. At higher altitudes, the collision rate drops
slowing the rate at which the translational temperature decreases.
The temperature of the local sublimation atmosphere is significantly increased when plasma heating is included
(Case 3). The temperature is high immediately above the surface of Loki but then decreases with altitude due to
radiation and adiabatic cooling and then rises again above 20 km where the plasma is able to penetrate (Figure 3C).
For all cases, the pressure over the hot spot does not equilibrate with the local sublimation atmospheric pressure
because dB-ATM < R. This is in qualitative agreement with Ingersoll et al. [4] who defined 2L Hπ= (when the
sticking coefficient is unity) as a horizontal averaging length scale over which local frost patches control the
pressure. In separate zero pressure gradient simulations, we found that as H is increased, the pressure became more
uniform over the hot spot. Since H and dB-ATM are directly proportional, this is the same as increasing the ballistic
length scale. When dB-ATM ≈ R, the pressure equilibrates over the hot spot.
ACKNOWLEDGMENTS
This work is supported by NASA PATM Grant #NNX08AE72G and OPR Grant #NNX08AQ49G. Simulations
were performed on supercomputers at the Texas Advanced Computing Center (TACC).
REFERENCES
1. J. Rathbun, J. Spencer, L. Tamppari, T. Martin, L. Barnard, and L. Travis, Icarus 169, 127-139 (2004). 2. F. Marchis and 11 colleagues, Icarus 176, 96-122 (2005). 3. A. Ingersoll, M. Summers, S. Schlipf, Icarus 64, 375-390 (1985). 4. A. Ingersoll, Icarus 81, 298-313 (1989). 5. M. Moreno, G. Schubert, J. Baumgardner, M. Kivelson, D. Paige, Icarus 93, 63-81 (1991). 6. D. Strobel, X. Shu, M. Summers, Icarus 111, 18-30 (1994). 7. M. Wong and R. Johnson, Icarus 115, 109-118 (1995). 8. M. Wong and R. Johnson, J. Geophysical Res. 101, 23243-23254 (1996a). 9. M. Wong and W. Smyth, Icarus 146, 60-74 (2000).
10. J. Austin and D. Goldstein, Icarus 148, 370-383 (2000). 11. C. Moore, D. Goldstein, P. Varghese, L. Trafton, B. Stewart, Icarus 201 585-597 (2009). 12. A. Walker, S. Gratiy, D. Goldstein, C. Moore, P. Varghese, L. Trafton, D. Levin, B. Stewart, Icarus 207, 409-432 (2010). 13. S. Gratiy, A. Walker, D. Levin, D. Goldstein, P. Varghese, L. Trafton, C. Moore, Icarus 207, 394-408 (2010). 14. G. Bird, Molecular Gas Dynamics and the Direct Simulation of Gas Flows. Oxford Univ. Press. Oxford, 1994. 15. J. Linker, M. Kivelson, R. Walker, 1988. Geophys. Res. Lett. 15, 1311–1314. 16. C. Moore, H. Deng, D. Goldstein, D. Levin, P. Varghese, L. Trafton, B. Stewart, A. Walker, Rarefied Gas Dynamics:
Proceedings of the 27th International Symposium, Monterey, CA. 17. S. Douté, B. Schmitt, R. Lopes-Gautier, R. Carlson, L. Soderblom, J. Shirley, Galileo NIMS Team, Icarus 149, 107-132 (2001).
18. D. Wagman, “Sublimation pressure and the enthalpy of SO2”. Chem. Thermodyn. Data Cent., Natl. Bur. of Stand. (1979). 19. S. Sandford and L.J. Allamandola, Icarus 106, 478-488 (1993).